Mercurial > repos > iuc > data_manager_bowtie_index_builder
comparison data_manager/bowtie_index_builder.py @ 5:08a3583826fa draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bowtie_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
| author | iuc | 
|---|---|
| date | Sun, 22 Nov 2020 12:44:59 +0000 | 
| parents | 86e9af693a33 | 
| children | 
   comparison
  equal
  deleted
  inserted
  replaced
| 4:86e9af693a33 | 5:08a3583826fa | 
|---|---|
| 11 CHUNK_SIZE = 2**20 | 11 CHUNK_SIZE = 2**20 | 
| 12 | 12 | 
| 13 DEFAULT_DATA_TABLE_NAME = "bowtie_indexes" | 13 DEFAULT_DATA_TABLE_NAME = "bowtie_indexes" | 
| 14 | 14 | 
| 15 | 15 | 
| 16 def get_id_name( params, dbkey, fasta_description=None): | 16 def get_id_name(params, dbkey, fasta_description=None): | 
| 17 # TODO: ensure sequence_id is unique and does not already appear in location file | 17 # TODO: ensure sequence_id is unique and does not already appear in location file | 
| 18 sequence_id = params['param_dict']['sequence_id'] | 18 sequence_id = params['param_dict']['sequence_id'] | 
| 19 if not sequence_id: | 19 if not sequence_id: | 
| 20 sequence_id = dbkey | 20 sequence_id = dbkey | 
| 21 | 21 | 
| 25 if not sequence_name: | 25 if not sequence_name: | 
| 26 sequence_name = dbkey | 26 sequence_name = dbkey | 
| 27 return sequence_id, sequence_name | 27 return sequence_id, sequence_name | 
| 28 | 28 | 
| 29 | 29 | 
| 30 def build_bowtie_index( data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME, color_space=False ): | 30 def build_bowtie_index(data_manager_dict, fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME, color_space=False): | 
| 31 # TODO: allow multiple FASTA input files | 31 # TODO: allow multiple FASTA input files | 
| 32 fasta_base_name = os.path.split( fasta_filename )[-1] | 32 fasta_base_name = os.path.split(fasta_filename)[-1] | 
| 33 sym_linked_fasta_filename = os.path.join( target_directory, fasta_base_name ) | 33 sym_linked_fasta_filename = os.path.join(target_directory, fasta_base_name) | 
| 34 os.symlink( fasta_filename, sym_linked_fasta_filename ) | 34 os.symlink(fasta_filename, sym_linked_fasta_filename) | 
| 35 args = [ 'bowtie-build' ] | 35 args = ['bowtie-build'] | 
| 36 if color_space: | 36 if color_space: | 
| 37 args.append( '-C' ) | 37 args.append('-C') | 
| 38 args.append( sym_linked_fasta_filename) | 38 args.append(sym_linked_fasta_filename) | 
| 39 args.append( fasta_base_name ) | 39 args.append(fasta_base_name) | 
| 40 args.append( sym_linked_fasta_filename ) | 40 args.append(sym_linked_fasta_filename) | 
| 41 tmp_stderr = tempfile.NamedTemporaryFile( prefix="tmp-data-manager-bowtie-index-builder-stderr" ) | 41 tmp_stderr = tempfile.NamedTemporaryFile(prefix="tmp-data-manager-bowtie-index-builder-stderr") | 
| 42 proc = subprocess.Popen( args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() ) | 42 proc = subprocess.Popen(args=args, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno()) | 
| 43 return_code = proc.wait() | 43 return_code = proc.wait() | 
| 44 if return_code: | 44 if return_code: | 
| 45 tmp_stderr.flush() | 45 tmp_stderr.flush() | 
| 46 tmp_stderr.seek(0) | 46 tmp_stderr.seek(0) | 
| 47 print("Error building index:", file=sys.stderr) | 47 print("Error building index:", file=sys.stderr) | 
| 48 while True: | 48 while True: | 
| 49 chunk = tmp_stderr.read( CHUNK_SIZE ) | 49 chunk = tmp_stderr.read(CHUNK_SIZE) | 
| 50 if not chunk: | 50 if not chunk: | 
| 51 break | 51 break | 
| 52 sys.stderr.write( chunk ) | 52 sys.stderr.write(chunk) | 
| 53 sys.exit( return_code ) | 53 sys.exit(return_code) | 
| 54 tmp_stderr.close() | 54 tmp_stderr.close() | 
| 55 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name ) | 55 data_table_entry = dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name) | 
| 56 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ) | 56 _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry) | 
| 57 | 57 | 
| 58 | 58 | 
| 59 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ): | 59 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): | 
| 60 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) | 60 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) | 
| 61 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] ) | 61 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) | 
| 62 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry ) | 62 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) | 
| 63 return data_manager_dict | 63 return data_manager_dict | 
| 64 | 64 | 
| 65 | 65 | 
| 66 def main(): | 66 def main(): | 
| 67 parser = optparse.OptionParser() | 67 parser = optparse.OptionParser() | 
| 68 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' ) | 68 parser.add_option('-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename') | 
| 69 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' ) | 69 parser.add_option('-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey') | 
| 70 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' ) | 70 parser.add_option('-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description') | 
| 71 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' ) | 71 parser.add_option('-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name') | 
| 72 parser.add_option( '-c', '--color_space', dest='color_space', action='store_true', default=False, help='color_space' ) | 72 parser.add_option('-c', '--color_space', dest='color_space', action='store_true', default=False, help='color_space') | 
| 73 (options, args) = parser.parse_args() | 73 (options, args) = parser.parse_args() | 
| 74 | 74 | 
| 75 filename = args[0] | 75 filename = args[0] | 
| 76 | 76 | 
| 77 params = json.loads( open( filename ).read() ) | 77 with open(filename) as fh: | 
| 78 target_directory = params[ 'output_data' ][0]['extra_files_path'] | 78 params = json.load(fh) | 
| 79 os.mkdir( target_directory ) | 79 target_directory = params['output_data'][0]['extra_files_path'] | 
| 80 os.mkdir(target_directory) | |
| 80 data_manager_dict = {} | 81 data_manager_dict = {} | 
| 81 | 82 | 
| 82 dbkey = options.fasta_dbkey | 83 dbkey = options.fasta_dbkey | 
| 83 | 84 | 
| 84 if dbkey in [ None, '', '?' ]: | 85 if dbkey in [None, '', '?']: | 
| 85 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( dbkey ) ) | 86 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (dbkey)) | 
| 86 | 87 | 
| 87 sequence_id, sequence_name = get_id_name( params, dbkey=dbkey, fasta_description=options.fasta_description ) | 88 sequence_id, sequence_name = get_id_name(params, dbkey=dbkey, fasta_description=options.fasta_description) | 
| 88 | 89 | 
| 89 # build the index | 90 # build the index | 
| 90 build_bowtie_index( data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME, color_space=options.color_space ) | 91 build_bowtie_index(data_manager_dict, options.fasta_filename, params, target_directory, dbkey, sequence_id, sequence_name, data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME, color_space=options.color_space) | 
| 91 | 92 | 
| 92 # save info to json file | 93 # save info to json file | 
| 93 open( filename, 'wb' ).write( json.dumps( data_manager_dict ) ) | 94 with open(filename, 'w') as fh: | 
| 95 json.dump(data_manager_dict, fh, sort_keys=True) | |
| 94 | 96 | 
| 95 | 97 | 
| 96 if __name__ == "__main__": | 98 if __name__ == "__main__": | 
| 97 main() | 99 main() | 
